Background: Model output can differ in many ways including
- variables
- attributes
- units
- levels
- file divisions for time-series (single time step in a file, single day, single month...)
- output grid resolution
- file names
- time format in netcDF (unlimited vs not unlimited)
Is it reasonable to suggest that any of these be standardized so that comparisons are easier. So that tools that analyze the data are easier to use?
Re: Reanalysis Model Output: Similarity of format and ...
The NASA Climate Model Data Service (CDS) and the NASA Center for Climate Simulation (NCCS) has initiated a new project is called the Collaborative REAnalysis Technical Environment – Intercomparison Project (CREATE-IP). The primary object of this project is to convert the existing reanalyses to a format consistent with the CMIP5 directives which includes standardized variable names, attributes, file names and time format. The data is distributed though the Earth System Grid Federation (ESGF) in netCDF CF compliant format at: https://www.earthsystemcog.org/projects/create-ip/. Each variable is stored separately and contains all time values. The horizontal resolution of each reanalysis is unchanged from the way it was produced but guidance is provided to assist with modification and all available model levels are included. Currently, the monthly variables provided are those with a one-to-one correspondence to the CMIP5 archive. High frequency output is now being processed to provide 1 hour, 3 hour and 6 hour data for a select set of 2D and 3D variables. In addition, plans are underway to publish the O-A, O-F terms and selected observations using the same formatting.